When online poker became a reality, it was only ever a matter of time before some programmers saw the potential to make some passive income.
We have always been taught that NL Hold’em and Pot Limit Omaha are not completely solvable because of the almost infinite decision tree. The GTO solvers we see today are proof that while a final strategy is not achievable, the silicon beast can get pretty damn close.
Spotting the IntrudersBots by their very nature play to a formula. This is the key to working out who they are and how to exploit them.
Expert level players have become extremely proficient at spotting these interlopers. While the rest of us mere mortals might think it almost impossible to work out what’s going on, there are now several strategies that are easily usable to spot a bot.
If you suspect a particular account, the simplest starting point is to observe the tables throughout your waking hours and check the suspected player. An account that is playing single block nine hour sessions is a candidate for suspicion, but so is one that plays in spurts at all times of the day, particularly if the session times cross into what would normally be sleeping hours.
Out of character stats are a major tell-tale. If you have several high-volume players in your player pool that all have the exact same stat that is wide of the norm, there is a good chance that further investigation might throw something else up.
This is typical of bot rings using the same program where the perpetrator has made accounts in friend’s names to maximise their profit.
Bet sizing is another area to keep an eye on. While many players will use preset buttons on the client or hotkeys from an application such as PartyCaption, they will periodically make an adjustment such as when going to max value.
Bots don’t tend to do this. They might have a few different bet sizes programmed in, but they will never move beyond this strategy. This reveal is at its strongest when the bet sizes are not the usual round numbers offered by the client.
Timing tells is the final check. We’ve all heard to keep an eye on what a particular opponents timing tells mean, but with a bot there is usually the exact same delay before clicking their decision.
Exploiting BotsMost of us wouldn’t know where to start when programming a bot to play poker. For a skilled programmer, though, it’s a straightforward process. The difficulty lies in covering all parts of the strategy, and this is an arduous task.
A lot of the programmers will stick with a basic strategy that is either easily exploited or just plain unbalanced.
For example, many players don’t have the stones to 5 bet bluff or just see it as preferable to fold and avoid the variance. For this reason many bots are known to have high 4 bet stats, but they also won’t call off wide. This is a highly profitable opportunity where you might even be able to exploit the bot by min five-betting.
You can even widen your three-bet range planning to five-bet wider than is normal.
Most bots also don’t adjust well to bet sizing, they only see the action and react the same no matter what the risk/reward is. If you are convinced you have identified a bot you should experiment with bluffing with smaller sizes.
If your success rate is the same then you have almost certain conformation your fears were correct and you can start to milk the unethical account holder.
You will also see a bot always make the same action against you in a particular spot. If it has access to many hands you played in a database it will see the exploit right away and automatically go after you.
This is a great opportunity if you can work out how it reacts to your next move. Maybe you are being aggressively raised on many turns or rivers because you correctly fold given the population tendencies. Once you start three-betting them with hands slightly weaker than you would normally go with see what happens. If you find the bot backs down every time then you have a profitable exploit.